When we hear about data centers “going green” and implementing sustainability measures, we usually think about reducing carbon emissions – and for good reason. Data centers use massive amounts of electricity, and the indirect carbon footprint involved with building and equipping them is impacted through the consumption and fabrication of raw materials throughout the entire upstream supply chain. According to the US Department of Energy, data centers account for roughly 2% of energy consumption in the US, and just over 1% of energy consumption globally. Although data centers have made progress in moving to renewable forms of energy, a substantial portion is still generated from burning fossil fuels. According to SuperMicro’s 2021 Data Centers and the Environment report, data centers are responsible for 2% of the world’s greenhouse gas emissions.
As important as reducing carbon emissions in data centers is, it is not the only sustainability issue that warrants attention. Another environmental factor that often gets overlooked when it comes to data centers’ ecological impact is gratuitous water usage.
Using Energy Means Consuming Water
Data centers use a lot of water. In 2015, it was estimated that a 15MW data center consumed roughly 130 million liters of water per year, roughly the same amount as a hospital with three buildings or a 100-acre almond orchard. And that’s just one data center. Multiply that by thousands of data centers all over the world, and you’re looking at billions of liters of water used to reject IT heat load, every year. According to some estimates, data centers consumed over 660 billion liters of water in 2020 alone.
Most of the water consumed by data centers is used to generate the electricity upon which these facilities depend. We all know that the majority of electricity we use is generated by fossil fuels. But few seem to realize that power plants use these fossil fuels to heat water, generating steam to turn the turbines that ultimately create the power. In other words, if you are using a significant amount of electricity, you are also de facto using a significant amount of water.
Power Usage Effectiveness
Data centers need power to operate. But not all of the power they use goes to running servers and other IT equipment. In fact, up to 40% of that power goes to cooling the data center. Servers generate heat and stop working when they get too hot, or when there are rapid temperature or humidity swings. Data centers therefore have to invest a lot of money and resources towards keeping servers cool enough to function. The most common type of cooling is air cooling, which relies on large air conditioning units that continuously blow cold air near the servers. (Otherwise known as the “Cold Aisle”.)
Needless to say, this is a very ineffective cooling method when you consider the environmental impact. To put this in perspective, a data center’s energy efficiency is calculated in Power Usage Effectiveness (PUE), which is a comparison of the amount of energy used by both the IT equipment and the facility, divided by the IT consumption figure. So, a PUE of 1 means that, in essence, no energy is wasted; all the energy goes to the IT equipment, none to the facility.
Air-cooled facilities can have a PUE as high 2 or even higher. This means that 50% of the energy used by the data center is wasted – that is, it goes to the facility itself, not IT services. Due to improvements in air cooling design and best practices, the average data center PUE is now 1.59 according to the Uptime Institute’s 2020 report. As we mentioned before, the electricity used to run these air cooling systems comes from power plants that not only emit a lot of CO2, but also use a significant amount of H2O.
Water Cooling Methods
Many data centers use chilled water as a cooling method. Data center owners and operators long ago turned to water for cooling because it is more efficient at heat removal than mechanical air conditioning. Water is used to cool servers by pumping it through cooling coils located adjacent, above, behind or near IT racks. Although in this case water is the heat transfer fluid, the heat exchange is still air-to-air.
Water can also be used to cool IT gear by pumping it through highly customized liquid cold plates that are mounted onto each microprocessor chip. This is known as direct liquid cooling (DLC). Using liquid cold plates to reject the heat generated by the dominant heat loads on the IT circuit board can successfully remove up to 70% of that heat. However, traditional air cooling is left to deal with the balance, which still burns a lot of carbon, consumes a lot of water, and leaves you with a complex series of piping, manifolds and tubing to deal with.
2-Phase Immersion Cooling Reduces Water Usage
Air and water cooling are both common methods of cooling in data centers, but they are far from ideal and are hitting the limits of their efficacy. The exponentially growing demand for data streaming brought on by the COVID-19 pandemic, combined with new data-intense technologies such as artificial intelligence (AI), IoT, 5G, high performance computing and cryptocurrency mining, require cooling capabilities that reach beyond the limits of air cooling and direct liquid cooling methods.
Simply put, above 300 watts per chip, air cooling becomes not only challenging but highly impractical and inefficient. There are GPU servers still being cooled with air, however these boxes can tower up to 10U high due to the extreme size of the heat sinks and fans required to reject the heat from the server. Further, a 10-12kW GPU server often sits isolated in its own rack with 80% of the other rack space units occupied by blanking panels and sometimes with an empty rack standing adjacent – this is known as “load spreading.” Most data centers are designed to cool with air at just 8-9kW per rack. A high-powered GPU server can necessitate limiting the IT load nearby, otherwise the air cooling system just can’t keep up with the demand.
2-phase immersion cooling is a game-changing technology that provides the cooling capability necessary to enable the use of advanced, data-heavy technologies while drastically reducing power consumption and, by extension, water usage.
With 2-phase immersion cooling, servers are immersed in a dielectric (nonconductive) fluid. The fluid absorbs the heat generated by the servers until it reaches its boiling point. The transition from liquid to gas is called a phase change, hence “phase one” of the process. This is a powerful mechanism because, as more heat energy enters the system, rather than raising the temperature of the fluid, it instead converts the fluid to gas. As a result, the electronics and surrounding fluid remain at a fairly constant temperature and the heat from the electronics can be transferred to an outside liquid loop without consuming appreciable amounts of energy.
The second phase change in 2-phase liquid immersion cooling is that of the gas changing back into a liquid thanks to a condensing coil placed just above the fluid. Condensed fluid collects on the coil then falls back into the tank, where the cycle is repeated continuously. No additional coolant pumps are required because the process is self-contained, and the fluid itself rarely has to be maintained or replaced. If IT gear is removed from the DataTanks™, the fluid evaporates quickly and cleanly, making the IT equipment very easy to service. While this is a concern with regard to potential fluid loss, LiquidStack has developed patented technology to drastically minimize this risk.
Another factor in water savings is that dry coolers, as the name implies, do not need to evaporate water to reject heat. A dry cooler is a closed-loop outside heat rejection system consisting of radiator coils and fans. Because 2-phase immersion cooling keeps primary loop fluid temperatures of 53°C or higher, dry coolers are able to easily and efficiently reject that IT heat load with free cooling even in the warmest, most humid, most challenging ambient environments in the world.
When data centers no longer rely on air conditioning or chilled water to cool their servers, they can operate at a much lower PUE. For example, by using 2-phase immersion cooling, data centers constructed in Hong Kong, the Republic of Georgia and Azerbaijan, some of the warmest climates in the world, operate at PUEs of 1.02 to 1.03, an order of magnitude better than a majority of air-cooled data centers.
In addition to limiting or eliminating direct water usage, 2-phase immersion cooling further saves on indirect water usage by reducing overall data center energy consumption. Reducing the consumption of local water resources also has a beneficial impact on local ecologies and communities.
Inadequate water supplies can lead to reduced food supplies and rising food costs. The need for fresh water is always increasing due to population growth, and using less of it reduces the potential effects of droughts and water shortages. Conserving water reduces the energy required to process and deliver it to homes, farms, and other businesses, further reducing carbon emissions and conserving fuel resources.
While the headline may be that 2-phase immersion cooling drastically reduces greenhouse gases, the reduction in water usage is also a valuable benefit to the planet and to any organization’s bottom line.